User's Guide

Table Of Contents
Data Collector Types
45 of 218
You can define your internal network by configuring sites. A site is a set of IP
masks that defines a well-known portion of your internal network. You can use
the World site to identify your entire internal network. If you have already
reserved certain IP address ranges for certain physical sites on your network, you
can create multiple sites that correspond to these reserved IP ranges. Multiple
sites can be created to identify different buildings, sites, or geographical areas of
your network. Any IP that matches any site is considered to be in-network. If you
define multiple sites, you will be able to analyze data broken down by site.
Data Collector Types
There are two kinds of data collectors used in ExtremeAnalytics.
l General Usage Collectors These are hourly and high-rate collectors that record
the top targets during an interval. Many types of targets and target-pairs are
supported.
l End-System Details Collector This is an hourly collector that attempts to capture
and record data for all in-network clients and servers that it detects. All traffic
collected is tagged with site, profile, device family, and other attributes.
Data from these collectors is stored separately in the database. The collector
data used in a report depends on the nature of the report. Higher-level
information, such as top applications during an hour, will be based on general
usage collector data, since it is relatively inexpensive to access. End-system
details data might be used when data for a specific client or server is needed, or
when the information requested is highly specific, for example, top applications
used by Android devices in the London site.
General Usage Collectors
General usage collectors collect data about all instances of a target for the
interval, and then record only the most significant targets (typically, the 100
most significant targets).
When the top targets are calculated for a collection interval, several different
statistics can be used as a basis for choosing the most significant entries. For
example, collectors can record the top applications based on bytes, and also
record the top applications based on number of clients. For each type of target
collected, there are different sets of bases used.